Glenn Entis, EA’s chief visual and technical officer, spoke today at a SIGGRAPH keynote entitled “Recent Accomplishments and Upcoming Challenges for Interactive Graphics in Video Games,” discussing the need for innovation and renewed “thrill-seeking” in the realm of interactive real-time graphics.
In the early days of CG, Entis reminisced, major breakthroughs were a regular phenomenon at SIGGRAPH, as innovations such as ray-tracing, fractals, and character animation burst onto the scene one year after another. At the time, these developments sent chills up the industry’s spine, but over time have become simple commodities, part of the “commoditization treadmill” that keeps us all in business.
“So how can we keep the thrills coming?” Entis pondered. The challenge, he said, is to find the new hot spots which excite both the creators and purveyors of graphical content. In this vein, Entis turned to the realm of real-time graphics – usually the awkward younger sibling of pre-rendered CG – which he sees as being the new frontier for the computer graphics community.
Challenges in Interactivity
A short recap of CG history reminded the audience that interactive graphics have trailed pre-rendered CG for decades, even back in 1982 when Tron was in theaters and Pac-Man
in the arcades. Games have come a long way since then – real-time rendered clips from the most recent Need for Speed
come remarkably close to the game’s pre-rendered cut scenes – but there’s a still a huge differential, in large part due to the enormous disproportion in frames rendered versus hardware cycles available.
That said, the interactive industry has an advantage, for while development technologies improve in both industries year after year, the technologies with which real-time graphics are rendered and displayed increase independently. “In short," said Entis, "we get double the Moore’s Law,” and as a result, the day of real-time rendering is fast upon us.
The question, is, “What creative opportunities are unique to interactive graphics?” Entis answered by examining upcoming challenges in three general areas of interest: game characters, the worlds they live in, and the tools used for their creation.
Characters: Beyond Uncanny Valley
Game characters are becoming increasingly anthropomorphic, said Entis, leading many developers into the perilous no-man’s land of “Uncanny Valley” – that unfortunate threshold upon which characters look too real, without the matching subtleties of behavior.
An interesting graph compared character modeling fidelity to motion fidelity, with Pac-Man
at the origin. When modeling fidelity is higher than motion fidelity, characters look better than they behave, and fall into “zombie territory.” Entis referred unsqueamishly to the film Final Fantasy: The Spirits Within as a classic example. Paradoxically, he said, many developers try to increase character realism by increasing the character’s polygon count – when in fact they need to either reduce the poly count, or increase motion fidelity.
On a similar graph, Entis compared character intelligence to motion fidelity, noting that characters must think smarter than their animation. Creating the “illusion of life” (referring to the famous book laying out Walt Disney’s 12 principles of animation) is the key to allowing users to believe what characters do, think, and feel, and moving beyond Uncanny Valley.
Entis demonstrated with clips from a real-time facial manipulation tool used at EA, in which animators can control facial expressions, eye movement, alertness, areas of awareness, etc, all with a simple set of abstracted graphical controls.
Environments: In Control of Nature
The first CG worlds, seen in movies such as Star Trek 2 and Tron, were awe-inspiring. Today, Entis said, we've successfully create one incredible world after another, and must go further to find new thrills in creating CG environments.
The trick here is that environments can’t merely look beautiful – they must behave beautifully, and be fully responsive to character and user input. Entis showed a demo from forthcoming EA-published game Crysis
– a 3D shooter in which every part of nature that can respond, does respond – and wowed the audience with simulated foliage and ricocheting bullets, dynamically generated ocean waves, volume rendered clouds, a dynamic HDR sky, and dozens of additional real-time graphical elements.
The challenge for environment creation, said Entis, is to gain control over the very forces of nature, and bring our CG worlds to life.
Tools: The New Content
Lastly, Entis discussed the role of tools in interactive entertainment. Usually, he said, we think of tools as being for content-makers – however the new trend is to understand tools as content, such as in The Sims, still the biggest selling game of all time.
The success of The Sims
, as well as websites such as YouTube and MySpace, have shown that so-called content-creation tools can in fact be as entertaining as the results of their use, opening up a new paradigm in interactive design. Here Entis demonstrated by showing off a character creation tool called “Virtual Me,” which allows for the creation of highly sophisticated avatars using only sliders and icons. He then showed a number of creature and vehicle building tools from Will Wright’s highly anticipated universe simulator, Spore
, in which the making of the creatures did indeed seem pretty darn fun.
Interactive: The Next Frontier
In short, Entis said, when the creators of entertainment get caught on the commoditization treadmill, the players feel it. Keeping the thrill in game development, and in graphics in general, means looking for the big new problems, finding innovative new ways to solve them, and then keeping ourselves excited by the new frontiers these solutions reveal.